Version 6 of this model is a merge of version 5 with RealVisXL by SG_161222 and a number of LORAs. One of the main goals is compatibility with the standard SDXL refiner, so it can be used as a drop-in replacement for the SDXL base model. It's very versatile and from my experience generates significantly better results than base SDXL, especially for characters. It's capable of some NSFW.
Description
Added PsyAnimated and new LORAs
FAQ
Comments (5)
Thanks for your work. Is it possible to release an inpainting version of this?
There is no sdxl-inpainting base model, so no. Inpainting models are made by taking the difference between the model and its base model and applying those differences to the base inpainting model.
@joeuser12 Thanks for info. Do you think an SDXL inpainting base will ever be released?
Can SDXL base models natively inpaint, or is that currently a lacking feature?
@Polygon They dont inpaint well in most scenarios. I have had awful results using base model.